SIM: A Search Engine by Correlating Scattered Data Sets for Cyber, Physical, and Social Systems

How to effectively locate a file in the emerging cyber, physical, and social systems is becoming a very important challenge due to the convergence of everyday life devices. A typical scenario goes like this. If you do not remember anything about a file (such as keywords or some text in the document) except some fragmented memories of what you are looking for, how can you locate the target in one of the many large folders in the system? A potential solution would be to use whatever you remember as a stepping stone along a path that leads to a small set of candidate targets. This paper proposes the design of a desktop search engine, Search Implicit Memories (SIM), which leverages both explicit and implicit memories. Explicit memories are employed as stepping stones to tap into the implicit memories, and locate the target in the implicit memories. The paper also proposes to monitor and correlate the existing data sets distributed across computer users' disk file systems to track user activities and construct correlation graphs. Experimental results show that SIM is both effective and lightweight. SIM can be used to complement existing keyword-based desktop search engines.